Received-SPF: pass (google.com: domain of linux-kernel+bounces-299-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Precedence: bulk
MIME-Version: 1.0
References: <20231208070754.3132339-1-stevensd@chromium.org> <9fb51bfc-576a-4c7b-92c2-c161e7f08ae7@redhat.com>
In-Reply-To: <9fb51bfc-576a-4c7b-92c2-c161e7f08ae7@redhat.com>
From: David Stevens <stevensd@chromium.org>
Date: Fri, 15 Dec 2023 10:00:41 +0900
Message-ID: <CAD=HUj6ZsZo_kxZazDDDUp61pEHupuf0Lf7WqkKBsMOA1CYcNQ@mail.gmail.com>
Subject: Re: [PATCH v2] virtio: Add support for no-reset virtio PCI PM
To: David Hildenbrand <david@redhat.com>
Cc: "Michael S . Tsirkin" <mst@redhat.com>, Jason Wang <jasowang@redhat.com>, 
	virtualization@lists.linux-foundation.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Thu, Dec 14, 2023 at 6:47=E2=80=AFPM David Hildenbrand <david@redhat.com=
> wrote:
>
> On 08.12.23 08:07, David Stevens wrote:
> > If a virtio_pci_device supports native PCI power management and has the
> > No_Soft_Reset bit set, then skip resetting and reinitializing the devic=
e
> > when suspending and restoring the device. This allows system-wide low
> > power states like s2idle to be used in systems with stateful virtio
> > devices that can't simply be re-initialized (e.g. virtio-fs).
> >
> > Signed-off-by: David Stevens <stevensd@chromium.org>
> > ---
> > v1 -> v2:
> >   - Check the No_Soft_Reset bit
> >
> >   drivers/virtio/virtio_pci_common.c | 34 +++++++++++++++++++++++++++++=
-
> >   1 file changed, 33 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/virtio/virtio_pci_common.c b/drivers/virtio/virtio=
_pci_common.c
> > index c2524a7207cf..3a95ecaf12dc 100644
> > --- a/drivers/virtio/virtio_pci_common.c
> > +++ b/drivers/virtio/virtio_pci_common.c
> > @@ -492,8 +492,40 @@ static int virtio_pci_restore(struct device *dev)
> >       return virtio_device_restore(&vp_dev->vdev);
> >   }
> >
> > +static bool vp_supports_pm_no_reset(struct device *dev)
> > +{
> > +     struct pci_dev *pci_dev =3D to_pci_dev(dev);
> > +     u16 pmcsr;
> > +
> > +     if (!pci_dev->pm_cap)
> > +             return false;
> > +
> > +     pci_read_config_word(pci_dev, pci_dev->pm_cap + PCI_PM_CTRL, &pmc=
sr);
> > +     if (PCI_POSSIBLE_ERROR(pmcsr)) {
> > +             dev_err(dev, "Unable to query pmcsr");
> > +             return false;
> > +     }
> > +
> > +     return pmcsr & PCI_PM_CTRL_NO_SOFT_RESET;
> > +}
> > +
> > +static int virtio_pci_suspend(struct device *dev)
> > +{
> > +     return vp_supports_pm_no_reset(dev) ? 0 : virtio_pci_freeze(dev);
> > +}
> > +
> > +static int virtio_pci_resume(struct device *dev)
> > +{
> > +     return vp_supports_pm_no_reset(dev) ? 0 : virtio_pci_restore(dev)=
;
> > +}
> > +
> >   static const struct dev_pm_ops virtio_pci_pm_ops =3D {
> > -     SET_SYSTEM_SLEEP_PM_OPS(virtio_pci_freeze, virtio_pci_restore)
> > +     .suspend =3D virtio_pci_suspend,
> > +     .resume =3D virtio_pci_resume,
> > +     .freeze =3D virtio_pci_freeze,
> > +     .thaw =3D virtio_pci_restore,
> > +     .poweroff =3D virtio_pci_freeze,
> > +     .restore =3D virtio_pci_restore,
> >   };
> >   #endif
> >
>
> Am I correct with my assumption that this will make s2idle work with virt=
io-mem-pci as well?
>
> Right now, all suspending is disabled, but really only s4/hibernate is pr=
oblematic.
>
> [root@vm-0 ~]# echo "s2idle" > /sys/power/mem_sleep
> [root@vm-0 ~]# echo "mem" > /sys/power/state
> [  101.822991] PM: suspend entry (s2idle)
> [  101.828978] Filesystems sync: 0.004 seconds
> [  101.831618] Freezing user space processes
> [  101.834569] Freezing user space processes completed (elapsed 0.001 sec=
onds)
> [  101.836915] OOM killer disabled.
> [  101.838072] Freezing remaining freezable tasks
> [  101.841054] Freezing remaining freezable tasks completed (elapsed 0.00=
1 seconds)
> [  101.843538] printk: Suspending console(s) (use no_console_suspend to d=
ebug)
> [  101.957676] virtio_mem virtio0: save/restore not supported.
> [  101.957683] virtio-pci 0000:00:02.0: PM: pci_pm_suspend(): virtio_pci_=
freeze+0x0/0x50 returns -1
> [  101.957702] virtio-pci 0000:00:02.0: PM: dpm_run_callback(): pci_pm_su=
spend+0x0/0x170 returns -1
> [  101.957718] virtio-pci 0000:00:02.0: PM: failed to suspend async: erro=
r -1

QEMU's virtio-pci devices don't advertise no_soft_reset, so this patch
won't affect vanilla QEMU. But if you add PCI_PM_CTRL_NO_SOFT_RESET to
the capability, then it should work. I'm working with crosvm, which
doesn't have virtio-mem implemented, but this patch makes virtio-fs
work with no extra kernel changes.

-David